I've had the same problem flying PA28s in S.Africa with 5400' elevation and 30 degrees OAT (ISA+25, work it out).
Most of the parameters affected by air density are smooth functions, so a moderate extrapolation would probably be accurate.
But as has been said, this isoutside the range of the factory test data, so you become the test pilot and, strictly speaking, your CofA is invalid (it usually says that a flight manual forms part of the certificate).
Obviously a lot of people work in the 'grey' bounds of the graphs. I'll leave the choice to you.